Dynamic tests on full-scale structures are of two main types: forced vibration tests (FVT), and ambient vibration tests (AVT). The FVT approach is further categorised into two types: free vibration tests, and steady-state forced vibration tests. For free vibration tests, an initial controlled excitation source is introduced to the structural system, and then removed, which causes the structure to perform damped free vi-brations. The response of the structure to the initial excitation is then recorded for analysis, but the input excitation force is not necessarily measured. The structure can be set in motion by either a sudden release from an initial displacement, by
imparting an initial velocity on the structure, or by imparting a force. Methods that have been successfully used to impart an initial excitation on tall buildings for free vibration tests are listed below;
• Mechanical shakers have been extensively used in previous vibration studies [9, 70, 111]. They are normally used for steady-state forced vibration tests, but are equally suited to imparting an initial force for a free vibration test. The input force from the mechanical shaker is repeated over a number of oscilla-tions to increase the vibration amplitude to the required level. The maximum amplitude of vibration achievable depends on the input force capability and the energy dissipation inherent to the structure. To achieve an accurate free vibration response for damping estimation, the force input from the mechani-cal shaker needs to be suddenly halted when the required vibration amplitude is attained. This requirement makes linear mechanical shakers [22] more suit-able than rotary mechanical shakers [69] for tall building dynamic testing.
Furthermore, rotary mechanical shakers become ineffectual at low vibration frequencies associated with tall buildings, because the accurate control of the shaker and the input forces generated both diminish with decreasing vibration frequency.
• Initial displacements of the structure can be achieved via an attached cable that is subsequently loaded to cause a desired level of displacement in the test structure [24]. In order to achieve an accurate free vibration result, a suitable release method is required to ensure the instantaneous and total relaxation of the tensile forces in the load cable. This method is more suitable for smaller buildings, as for taller buildings the length of cable and the angle of the applied force may be prohibitive.
• Synchronised movement of one or more humans at the natural frequency of the test structure has been shown to achieve vibration amplitudes suitable for free oscillation tests [59, 72, 166]. Best results are achieved if all involved in generating the excitation force are pushing in unison against a common structural wall. This method is essentially the same as using a mechanical shaker, except that measurement of the input force would require much more effort. Like a mechanical shaker, a number of input pushes can be made in succession to increase the vibration amplitude. The main benefits of this method are simplicity, cost effectiveness, and the ability to instantaneously halt the excitation force.
• A construction crane attached to a building can be used as an excitation source by lowering a mass and suddenly breaking the fall [42, 59]. Multiple fall and
braking cycles can be made in succession to increase the vibration amplitudes, and the boom can be positioned at different locations relative to the building axes to target particular vibration modes.
If damping ratios are required from free vibration tests, it is important that am-bient excitations are minimised during testing so that the applied initial excitation is the only significant action on the test structure. Examples of ambient excitations for tall buildings include wind loading, earthquake tremors, construction crane ac-tivity, and elevator car movements. Minimising these sources of unmeasured force inputs will ensure errors for damping values are within acceptable limits.
The second type of FVT, the steady-state forced vibration test, requires sinu-soidal forces of varying frequency and amplitude to be applied to the test structure.
Both the input force and output response are recorded, and combined to generate resonance curves or frequency response functions [74]. The dynamic characteristics of the test structure can then be extracted from the resonance curves. A mechan-ical shaker is used to apply the sinusoidal forces, and like the free vibration tests, the minimisation of the unmeasured ambient excitation forces is important to avoid excessive errors in results. This method has been successfully used for the dynamic testing of numerous tall buildings [9, 71, 112, 154].
Dynamic tests using the AVT approach have also been extensively used in pre-vious studies [17, 33, 173]. Like the free vibration test method, the input excitation force is not measured. However, the output from ambient vibration tests are not free vibration responses from a single excitation. Rather, the output response is generated by multiple excitation forces acting on the test structure, which vary with time and have various degrees of spacial correlation.
Ambient loading has the advantage of exciting multiple natural modes of vibra-tion that can be recorded during a single test, and viewed individually using digital techniques while post processing the vibration signals. In comparison to FVT, the forced loading of large civil engineering structures has a number of disadvantages.
Development and logistical disadvantages are usually the most apparent when con-sidering forced loading. Development of mechanical shakers used to conduct forced loading is often time consuming and costly in comparison to ambient loading, which requires no development of loading equipment.
Conducting forced loading is often accompanied by logistical problems associated with the transportation and installation of the loading equipment. Depending on the size of the structure to be tested, the loading equipment can be extremely large and heavy, and therefore difficult to transport, in addition to causing potential disruptions to the regular operation of the structure while conducting the tests.
Another disadvantage of forced loading arises from requiring almost complete control of the various input loads on the structure. This results in forced loading tests being
conducted when ambient loads are minimised — calm wind conditions for example.
Control of ambient loads is practically impossible, which means forced loading tests are at best scheduled when ambient loads are expected to be minimised.
Previous research [121, 154, 175] has concluded that dynamic testing of buildings using ambient excitation can provide reliable estimates of natural frequencies and mode shapes. However, until more recently the estimation of structural damping ratios from ambient excitation tests was particularly unreliable. In contrast, under ideal forced excitation tests, structural damping ratios can be reliably determined.
The primary reason for this is under ideal conditions the input force can be accurately determined when conducting a FVT. The known input force is used along with the response output to determine the dynamic properties. Furthermore, FVT can also force a particular mode to be excited with-out other modes ocurring simultaneously. For AVT there is also the issue of separating structural damping and aerodynamic damping components from AVT test results.
More recent techniques [14, 17, 127, 136] for post processing ambient excitation test results have improved the structural damping ratio estimates, however most techniques are more computationally intensive when compared with the processing requirements of forced excitation test data.
Vibration amplitudes generated by ambient excitation forces are usually orders of magnitude less than those generated by forced excitation. Unless a broad spectrum of vibration amplitudes are experienced during the testing phase, the results will only be valid for a small subset of potential loading conditions. This is a weakness of ambient vibration testing, particularly when considering the influence of response amplitude on natural frequency and structural damping ratios. It has been ob-served that as vibration amplitude increases, the natural frequencies decrease and the structural damping ratios increase [80, 87, 165]. This effect is reflected in many building design codes, which specify larger values for structural damping ratios for ultimate load cases compared with the values for serviceability load cases [151].
2.4.1 Dynamic Testing during Construction
The previous sections highlight the large amount of data and knowledge obtained on the full-scale dynamic characteristics of completed structures. Much of this has been due to research focused on earthquake engineering of civil structures. Very few previous research efforts have investigated vibration testing of partially completed structures during construction, either as a means of understanding the structural mechanisms that influence the dynamic characteristics, or for enhancing the pre-diction of dynamic characteristics. Compared with testing completed structures, testing during construction presents more obstacles to achieving optimum outcomes.
It is difficult to control the construction of a building to maximise the information
from vibration tests during construction. The schedule of the builder, who is under pressure to meet cost and time budgets, governs how the structure is built. In most cases, vibration testing during construction can only aim to conduct tests at the most opportunistic stages of construction, in order to observe the effects of structural changes on the dynamic characteristics.
Tall building construction is a particularly good example, because the construc-tion schedule tends to progress based on levels, as opposed to structural elements.
It would be advantageous to conduct vibration tests when certain parts of the struc-ture are completed, for example at the completion of the shear walls, columns, floor slabs, facade, and internal non-structural walls. However, tall building construction tends to overlap all of these parts. For example, the installation of the facade is generally commenced once the structure has been completed far enough ahead, but not entirely completed, to avoid clashes between construction processes. In this arrangement, the facade and structure are built simultaneously, with the facade trailing until the structure reaches completion.
Complete control over the construction of the structure would be ideal for under-standing the structural mechanisms and the influence of both structural and non-structural elements, but this is not practical or economical for tall buildings. Despite not having complete control — to add or remove structural and non-structural ele-ments at any time — during construction, determining the dynamic characteristics at partially completed states can provide insight into the structural mechanisms that influence the dynamic characteristics.
Due to the construction schedule mentioned above, it is unlikely that one study will ever provide comprehensive understanding of tall building dynamics. Each study of a building during construction offers a unique opportunity to further the understanding, particularly if the construction schedule differs from previous studies.
Dynamic tests of partially completed buildings, or buildings undergoing alter-ations, are not a recent occurence. The earlier studies conducted in the 1930’s were mostly confined to steel frame buildings of under fifteen storeys [23], which were the dominant form of construction at that time. For this type of building, the main focus appeared to be determining the influence of curtain walls, partition walls, and concrete encasement of the steel frame [8, 24, 145]. Furthermore, the results focused attention on the natural periods of oscillation, and little information was provided on the damping ratios. More recent studies have since included damping ratio estimates [118].
The most comprehensive study for steel frame structures is perhaps the Card-ington steel frame building [49]. This steel frame structure comprised eight storeys and was designed to represent an office building. The entire structure was con-structed within a laboratory, which allowed for practically complete control of the
input excitation. Furthermore, the construction was undertaken in discrete stages to accurately determine the influence of each stage on the dynamic characteristics.
This controlled laboratory experiment is at the extreme end of full-scale testing, and generally not a viable option for larger structures.
Taller buildings have also been tested during construction and alterations, and of particular interest to this research are those [16, 42, 141, 178] that use a reinforced concrete core as the primary lateral load resisting structure. These few studies represent the current breadth of research that use full-scale testing of tall buildings during construction to improve the estimation of dynamic characteristics.
2.4.2 Natural Frequency during Construction
A majority of the research on the estimation of natural frequencies during construc-tion has focused on the influence of non-structural components, such as internal par-titions and facades. Shorter buildings have been more successful test cases because the reduced number of levels, compared with taller buildings, means the primary lateral system is more likely to be completed prior to the construction of the facade and non-structural partitions. Therefore, the changes can be observed in discrete stages.
For taller buildings during construction, the ability to observe the influence of the facade and internal partitions is limited to the final stages of construction. This is due to the main structure being completed while the remaining fit-out and facade are conducted at lower levels, thus creating a stage at the end of construction where only the facade and fit-out are changing. The observation of this stage in previous studies of reinforced concrete core buildings has found the facade and partitions did not significantly influence the dynamic characteristics [16, 42, 178]. This is a similar result to that obtained for a steel frame tower of 24 storeys, which found the cladding had little influence on the first modes, but tended to increase the natural frequencies of the higher modes of vibration [118]. Regardless of the initial aim of the previous studies, all the results have reported a decrease in natural frequency with increasing height.
A relatively complete study of natural frequencies for a tall building during construction was conducted on a reinforced concrete tower in Vancouver [141]. The tower included 30 storeys, with 85 m height above ground and 9.4 m of basement below ground. Using ambient vibration measurements, the natural frequencies of the fundamental vibration modes — two sway and one torsion — and the second and third set of higher modes was estimated. The results found the spacing between the fundamental and higher order modes decreased as the building height increased. For the sway modes, the ratios between the higher modes and the fundamental mode was considerably higher, by as much as two times, than the ratios for an idealised shear
beam model in Equation (2.2). The ratios for the torsional modes were generally within 10% of the idealised shear beam model.
2.4.3 Damping Ratios during Construction
As for the investigation of natural frequencies during construction, previous research has also focused on determining the influence of non-structural components on the damping ratios. A common obstacle shared amongst the previous research is the difficulty in obtaining a consistent set of accurate estimates for the damping ratios during construction [42, 141]. This is understandable since the estimation of damp-ing ratios is very sensitive to the loaddamp-ing conditions. And this fact is not avoided if a mechanical shaker is used to excite the structure, as ambient vibrations can still influence the results.
Regardless of the incomplete results sets for damping ratios, potentially useful results have been reported. For a 24 storey steel frame tower, the facade was found to increase the damping ratios, with the effect being more pronounced for the tor-sional modes [118]. Forced vibration tests of a 94 m high apartment building with reinforced concrete core found the damping ratios to be more dependent on the stage of fit-out in the lower levels, as opposed to the fit-out in the higher levels [42].